Identification of Functional Segments within the β2I-domain of Integrin αMβ2

The αMβ2integrin plays an important role in leukocyte biology through its interactions with a diverse set of ligands. Efficient ligand binding requires the involvement of both the αM and β2 subunits. Past ligand binding studies have focused mainly on the αM subunit, with the β2subunit being largely unexplored. Therefore, in this study we conducted homolog-scanning mutagenesis on the I-domain (residues 125–385) within the β2 subunit. We identified four noncontiguous sequences (Arg144–Lys148, Gln199–Ala203, Leu225–Leu230, and Gly305–His309) that are critical for fibrinogen and C3bi binding to αMβ2. Molecular modeling revealed that these four sequences reside within a narrow region on the surface of the β2I-domain, in close proximity to three potential cation-binding sites. Among these sequences, Gln199–Ala203, Leu225–Leu230, and Gly305–His309 are important for the binding of both ligands, whereas Arg144–Lys148 is more critical for fibrinogen than for C3bi binding. These sequences within the β2I-domain are directly involved in ligand binding, since 1) switching these segments to their corresponding β1 sequences destroyed ligand binding; 2) loss of function was not due to a nonspecific gross conformational change, since the defective αMβ2 mutants reacted well with a panel of conformation-dependent mAbs; 3) mutation of these functional sequences did not effect Ca2+binding; and 4) synthetic peptides corresponding to sequences Gln199–Ala203 and Gly305–His309 blocked ligand binding to αMβ2, and the peptides interacted directly with fibrinogen and C3bi. Given the similarity among all integrin β subunits, our results may help us to understand the underlying mechanism of integrin-ligand interactions in general.

Site-directed Mutagenesis and Establishment of Stable Cell Lines-The detailed procedures for homolog-scanning mutagenesis and establishment of stable cell lines expressing wild type and the 16 mutants of ␣ M ␤ 2 in human kidney 293 cells have been published (13). Similar procedures were used to generate an additional mutant, ␣ M ␤ 2 (Leu 225 -Leu 230 ), The mutagenic primer used to create this mutant was 5Ј-GA-GGTCGGGAAGCAGAGCGTCTCCAGGAACAGGGATGC-ACCCGAG-GGT-3Ј, which switched the 225 LISGNL segment in the ␤ 2 I-domain to the corresponding sequence (SVSRNR) in the ␤ 3 I-domain. To obtain cell lines that express equivalent receptor numbers as wild-type ␣ M ␤ 2 , each mutant cell line was subcloned by cell sorting using an ␣ M -specific mAb 2LPM19c. Up to 20 colonies were picked and analyzed for integrin expression by FACS analysis. Cells expressing similar levels of receptor to those expressing wild-type ␣ M ␤ 2 were selected and subcloned. To exclude the possibility of subcloning artifacts, all of our studies have been repeated using the original pool for every mutant.
FACS Analysis-A total of 10 6 cells expressing wild-type or mutant ␣ M ␤ 2 in Hanks' balanced salt solution (HBSS) containing 1 mM Mg 2ϩ and 1 mM Ca 2ϩ were incubated with 1 g of mAb for 30 min at 4°C. A subtype-matched mouse IgG served as a control. After washing with PBS, cells were mixed with fluorescein isothiocyanate-labeled goat antimouse IgG(HϩL) F(abЈ) 2 fragment (1:20 dilution) (Zymed Laboratory) and kept at 4°C for another 30 min. Cells were then washed with PBS and resuspended in 500 l of DPBS. FACS analysis was performed using FACScan (Becton-Dickinson), counting 10,000 events. Mean fluorescence intensities were quantified using the FACScan program, and the values were used to compare ␣ M ␤ 2 expression levels for the subclones of each mutant or the reactivity of different ␤ 2 mutants with various ␤ 2 -specific mAbs.
Ligand Binding Assays for ␣ M ␤ 2 -The ligand binding activity of the ␤ 2 mutants was assessed using two classic ␣ M ␤ 2 ligands, C3bi and Fg, according to our published methods (11,13). The adhesion of ␣ M ␤ 2expressing cells to Fg was assessed using the recombinant ␥-module, which is the principle binding site for ␣ M ␤ 2 (14). 24-Well polystyrene plates were coated with the ␥-module (10 g/ml). After blocking with 400 l of 0.05% polyvinylpyrrolidone in DPBS, a total of 2 ϫ 10 6 cells in HBSS containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ in the presence or absence of 2 mg/ml synthetic peptides or 20 g/ml mAb 44a (against the ␣ M subunit) were added to each well and incubated at 37°C for 20 min. Unbound cells were removed by three washes with DPBS, and adherent cells were quantified by cell-associated acid phosphatase as described previously (11).
Solid Phase Binding Assays-To test the interaction between the identified sequences of the ␤ 2 I-domain and the ␣ M ␤ 2 ligand Fg, 96-well microtiter plates (Immulon 4BX; Dynex Technologies Inc., Chantilly, VA) were coated with different synthetic peptides at 2 mg/ml overnight at 4°C and postcoated with 3% BSA for 2 h at room temperature. The Fg ␥-module (10 g/ml) in TBS (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 0.05% Tween 20) was added to the wells and incubated for 2 h at 22°C. After washing with TBS, bound ␥-module was detected using a sheep anti-␥-module antibody, a donkey antisheep IgG conjugated to horseradish peroxidase, and the horseradish peroxidase substrate, 3,3Ј,5,5Ј-tetramethylbenzidine (KPL, Gaithersburg, MD). To examine the interaction between the ␤ 2 I-domain peptides and C3bi, 80 l of the synthetic peptides (2 mg/ml in DPBS) were coated onto 24-well plates overnight at 4°C. After blocking with 1% BSA for 1 h at 22°C, biotinylated EC3bi in 300 l of HBSS containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ were added. After incubating for 1 h at 37°C, nonadherent EC3bi was removed by washing. The plate was then fixed with 2% paraformaldehyde and blocked with 2% BSA. C3bi binding was measured with a conjugate of avidin-alkaline phosphatase, and the amount of bound EC3bi was determined by reaction with p-nitrophenyl phosphate, measuring the absorbance at 405 nm.

RESULTS
Ligand Binding by the Homolog-scanning Mutants of ␣ M ␤ 2 -Previously, using homolog-scanning mutagenesis, we systematically mutated residues that reside on the surface of the ␤ 2 I-domain and mapped the epitopes of several function-blocking mAbs. We demonstrated that all 16 ␤ 2 I-domain mutants, when transfected together with wild-type ␣ M , were expressed at wild-type levels on the cell surface. In addition, five homologscanning mutants, which contain individual substitutions of the segments Leu 154 -Glu 159 , Pro 192 -Glu 197 , Asn 213 -Glu 220 , Glu 344 -Asp 348 , and His 354 -Asn 358 that constitute the epitopes for nine function-blocking mAbs, were studied for ligand binding and found to bind to both ␣ M ␤ 2 substrates (Fg and C3bi) (13). Within the ␤ 3 I-domain, 211 SVSRNRDAPEGG 222 has been identified as a critical sequence for ligand recognition by ␤ 3 integrins (15). Most recently, crystallographic studies of the RGD-␣ V ␤ 3 complex have confirmed that the three ␤ 3 residues, 214 RNR, were in direct contact with RGD (16). Therefore, to see if this sequence is also important in ligand binding by ␣ M ␤ 2 , we constructed a new ␤ 2 I-domain mutant by switching the corresponding sequence ( 225 LISGNL 230 ) within the ␤ 2 I-domain to its homolog (SVSRNR) within ␤ 3 , since this region is well conserved between ␤ 2 and ␤ 1 . To access the functionality of these ␤ 2 mutants, we conducted a ligand binding experiment following our established assays (13). Two representative ligands of ␣ M ␤ 2 , the ␥-module of Fg and C3bi, were chosen in this study, because they recognize overlapping but not identical binding sites within the receptor (11). As shown in Fig. 1A, most of the mutants, including the five mutants (␣ M ␤ 2 (Leu 154 -Glu 159 ), . Two mutant cell lines, ␣ M ␤ 2 (Arg 144 -Lys 148 ) and ␣ M ␤ 2 (Gly 305 -His 309 ), had modest loss of cell adhesion to the ␥-module (ϳ2.5-fold). The most dramatic changes in ligand binding (Ͼ10-fold different from the wildtype receptor) were observed for the following two ␤ 2 I-domain mutants: ␣ M ␤ 2 (Gln 199 -Ala 203 ) and ␣ M ␤ 2 (Leu 225 -Leu 230 ). The specificity of ligand binding to each of the mutants that retained activity was verified by blocking experiments using an ␣ M -specific mAb, 44a. Furthermore, we verified that the wildtype receptor, ␣ M ␤ 2 (Leu 225 -Leu 230 ), and the other 16 ␣ M ␤ 2 mutants had similar surface expression levels (Ͻ2-fold), as assessed by FACS analysis using ␣ M -specific mAb 44. In addition, to exclude possible clonal effects resulting in defective ligand binding, we repeated the above adhesion experiments using two additional independent clones of each defective mutant. These data suggest that the Fg binding pocket is composed of four segments (Arg 144 -Lys 148 , Gln 199 -Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 ) within the ␤ 2 I-domain.
Next, we evaluated the effects of the ␤ 2 I-domain mutations on C3bi binding. As shown in Fig. 1B, the mutant with more than 10-fold decrease in C3bi binding activity was ␣ M ␤ 2 (Leu 225 -Leu 230 ) (in comparison with wild-type binding). , and ␣ M ␤ 2 (His 371 -Lys 379 ) interacted well with C3bi (less than 1.5-fold difference). As a control, mock-transfected cells had minimal C3bi binding. In addition, C3bi binding by the wild type and the functional mutants was blocked by the addition of 1 mM EDTA. These data suggest that three noncontiguous segments, Gln 199 -Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 , are critical for C3bi binding. Combining the above data, we concluded that these two ligands recognize a common region within the ␤ 2 I-domain (Gln 199 -Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 ), with one exception; segment Arg 144 -Lys 148 is a critical sequence for ␣ M ␤ 2 -mediated cell adhesion to Fg but not for C3bi binding (13). Thus, the ligand binding sites with the ␤ 2 I-domain for Fg and C3bi are overlapping but not identical.
The ␤ 2 I-domain Mutants Exhibited Correct Conformations-To exclude the possibility that the defective ligand binding by the ␤ 2 I-domain mutants resulted from gross alterations of the structures of the mutant ␣ M ␤ 2 receptors, we conducted a number of experiments. First, as reported previously, all mutant ␣ M ␤ 2 receptors were expressed on the cell surface. Surface labeling and immunoprecipitation experiments conducted on cells expressing these ␣ M ␤ 2 mutants showed that all mutant ␣ M ␤ 2 had similar molecular sizes to that of the wild-type receptor (13). Second, the four defective mutants could be recognized by a panel of conformation-dependent ␤ 2 -specific mAbs, including MEM48, 6.7, 7E4, TS1/18, and CLB54, judged by FACS analysis (13). In particular, the epitopes of TS1/18 and CLB54 are composed of at least three noncontiguous segments within the ␤ 2 I-domain and therefore could only recognize correctly folded ␣ M ␤ 2 (13,17). Third, using mAbs (YFC 118.3 and CLB54) that depend on both a correct conformation and a functional Ca 2ϩ binding site for their optimal recognitions of the ␤ 2 I-domain (13), we probed the integrity of the Ca 2ϩ -binding site within the ␤ 2 subunit for the above four defective mutants. Antibody binding to the four defective mutants was determined by FACS analysis in the presence of different concentrations of Ca 2ϩ . Since mutation of segment Arg 144 -Lys 148 destroyed the epitope for mAb YFC118.3 (13), we used mAb CLB54 for evaluating the Ca 2ϩ binding affinity of mutant ␣ M ␤ 2 (Arg 144 -Lys 148 ). Like mAb YFC118.3, mAb CLB54 recog- , and ␣ M ␤ 2 (Gly 305 -His 309 ) (OE) was measured using a conformationdependent mAb CLB54 or mAb YFC118.3, both of which require an intact Ca 2ϩ binding site within the ␤ 2 subunit for optimal recognition of the ␤ 2 I-domain (13). Bound antibody was determined by FACS analysis, and the mean fluorescence intensity was calculated using the FACScan program. The titration data can be fitted into the singlebinding site model using the nonlinear regression program provided by SigmaPlot (SPSS Science, Chicago, IL). To compare the titration data of the four defective mutants, which had similar K d values to the wild-type receptor (ϳ100 M), the amount of bound mAb at 1000 M Ca 2ϩ , after subtraction of background, was taken as 100%. Data are the means Ϯ S.D. of three independent experiments.

FIG. 1. Ligand binding to the ␤ 2 Idomain homolog-scanning mutants.
A, Fg adhesion. A total of 2 ϫ 10 6 ␣ M ␤ 2expressing cells were added to 24-well nontissue culture polystyrene plates, which were precoated with recombinant ␥-module (10 g/ml) and subsequently blocked with 0.05% polyvinylpyrrolidone in DPBS. After incubation at 37°C for 20 min, the unbound cells were removed by three washes with DPBS, and the adherent cells were quantified by cell-associated acid phosphatase. The number of adherent cells expressing wild-type ␣ M ␤ 2 was taken as 1.0. Specificity was demonstrated by the addition of an ␣ M -specific function-blocking mAb, 44a. *, a value of less than 1%. Data shown are the means Ϯ S.D. of three independent experiments. B, C3bi binding. Biotinylated EC3bi (2 ϫ 10 7 ) were added to 2 ϫ 10 5 cells expressing ␣ M ␤ 2 , which had been preseeded onto polylysine-coated 24-well plates. After 60 min at 37°C, the number of bound EC3bi was determined using avidin-alkaline phosphatase and p-nitrophenyl phosphate, measuring the absorbance at 405 nm. The value for wild-type ␣ M ␤ 2 was taken as 1.0. Specificity was demonstrated by the addition of EDTA (1 mM). *, a value of less than 1%. Data are the means Ϯ S.D. of 3-6 independent experiments. Note that the Fg adhesion and C3bi binding data for the six mutants nizes ␣ M ␤ 2 in a cation-dependent manner. Similar K d values for Ca 2ϩ binding to the wild-type receptor were obtained using mAbs YFC118.3 and CLB54 (data not shown). As a representative, Fig. 2 shows a Ca 2ϩ titration experiment for the defective mutant ␣ M ␤ 2 (Leu 225 -Leu 230 ), probed with mAb YFC118.3. For all of the defective mutants and the wild-type receptor, antibody binding increased with increasing Ca 2ϩ and saturated above 1 mM added Ca 2ϩ . These data could be fitted to a singlebinding site model. The estimated K d of this Ca 2ϩ -binding site for wild type was 73.5 M, which is very close to the K d value of 105 M we reported earlier (13). The K d values obtained for the four defective mutants were 107 M for ␣ M ␤ 2 (Arg 144 -Lys 148 ), 70 M for ␣ M ␤ 2 (Gln 199 -Ala 203 ), 85.4 M for ␣ M ␤ 2 (Leu 225 -Leu 230 ), and 51 M for ␣ M ␤ 2 (Gly 305 -His 309 ), demonstrating that the high affinity Ca 2ϩ binding site within the ␤ 2 I-domain was intact for all four defective ␤ 2 I-domain mutants. Thus, we concluded that these four function-destroying ␤ 2 I-domain mutations did not significantly alter the gross conformation of the ␣ M ␤ 2 receptor; nor did they change the affinity of the Ca 2ϩ binding site within the ␤ 2 subunit. Therefore, the loss of ligand binding function by these four defective ␤ 2 mutants was most likely due to a direct perturbation of the ligand binding site.
Synthetic Peptides of the ␤ 2 I-domain Blocked Ligand Binding to ␣ M ␤ 2 -To further verify that the four identified sequences were directly involved in ligand binding, we prepared four synthetic peptides, P1 (CDLSYSMLDDLR), P6 (CQPP-FAFRHVLK), P11 (CSVGLAHKLAE), and P17 (KQLISGN-LDAPEGGLD), and tested their capacity to compete with wildtype ␣ M ␤ 2 for binding to Fg. The choice of these four peptides was based on the above mutagenesis data plus analysis of the predicted ligand binding site by molecular modeling. As shown in Fig. 3, among these peptides, only peptide P6 was effective in blocking cell adhesion to the ␥-module by the wild-type receptor, reducing the number of adherent cells by 5-fold. Peptide P11 exhibited small but significant inhibition (1.5-fold), whereas peptides P1 and P17 and two control peptides (KYGRGDS and a scrambled peptide CFKLPHVARPQF (sP6)) had no effect. These data suggest that the ␤ 2 I-domain peptides were capable of interfering with the interactions between ␣ M ␤ 2 and its ligand Fg, possibly by binding directly to Fg.
Direct Interaction of the ␤ 2 I-domain Peptides with Fg and C3bi-In these experiments, we evaluated the capacity of the ␤ 2 I-domain peptides to directly interact with Fg and C3bi. For peptide-Fg interaction, we coated the ␤ 2 I-domain peptides onto 96-well microtiter plates, and the binding of the Fg ␥-module to the peptides was assessed. As shown in Fig. 4A, only peptide P6 bound the ␥-module effectively. The other three peptides as well as a scrambled control peptide (sP6) did not bind the ␥-module. To see whether the synthetic peptides could directly bind C3bi, we coated these ␤ 2 I-domain peptides onto microtiter plates. Binding of the biotinylated EC3bi to these peptides was then determined. As shown in Fig. 4B, two peptides, P6 and P11, bound C3bi strongly, whereas peptides P1 and P17 had no detectable binding. Similarly, all controls were negative. DISCUSSION In this work, we have studied the ligand binding site within the ␤ 2 I-domain using the homolog-scanning mutagenesis approach. We report three major findings. 1) Fg and C3bi recognize overlapping but nonidentical binding sites within the ␤ 2 Idomain. The ligand binding site is composed of four noncontiguous sequences: Arg 144 -Lys 148 , Gln 199 -Ala 203 , Direct interaction between the synthetic peptides and the Fg ␥-module was measured by enzymelinked immunosorbent assay. Briefly, 96-well microtiter plates were coated with 50 l of synthetic peptides (2 mg/ml) overnight and then blocked with 3% BSA. The Fg ␥-module (10 g/ml) was added to the wells and incubated for 2 h at 22°C. After washing, bound ␥-module was detected with a sheep anti-␥-module antibody and a donkey antisheep IgG conjugated to horseradish peroxidase. The amount of bound Fg ␥-module was measured by reaction with a horseradish peroxidase substrate, 3,3Ј,5,5Ј-tetramethylbenzidine. B, C3bi binding. Eighty microliters of 2 mg/ml peptides was used to coat the center of each well of a 24-well nontissue culture plate. The plate was washed with PBS and blocked with 1% BSA. Then biotinylated EC3bi in HBSS containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ were added and incubated for 1 h at 37°C. After washing, fixation, and blocking, the amount of bound EC3bi was determined using an avidin-alkaline phosphatase conjugate and pnitrophenyl phosphate as the substrate, measuring the absorbance at 405 nm. Data shown are the means Ϯ S.D. of three independent experiments.
Leu 225 -Leu 230 , and Gly 305 -His 309 . 2) All four defective homolog-scanning mutants exhibited correct conformation, evidenced by their reactivity toward mAbs that are conformationand cation-dependent. Most importantly, these four defective mutants all possessed an intact and high affinity Ca 2ϩ binding site within their respective ␤ 2 subunits. 3) The corresponding synthetic peptides (Gln 199 -Ala 203 and Gly 305 -His 309 ) were capable of competing with the intact receptor for Fg binding, and, when coated on microtiter wells, both interacted directly with Fg and C3bi.
The ligand binding sites within the ␤ 2 I-domain were identified initially using a homolog-scanning mutagenesis approach (12), which we successfully used in our previous studies (14,18,19). This approach is based on the sequence similarity (74%) but functional disparity between the ␤ 1 and ␤ 2 integrins. To preserve the gross structure of the ␤ 2 I-domain, none of the mutations in this study involved residues that either played an essential role in protein folding, or provided a coordination sites for cations, such as Asp 134 , Ser 136 , Ser 138 , and Glu 234 or Asp 264 (equivalent to Thr 209 and Asp 242 of the ␣ M I-domain, respectively) (20). Our earlier studies strongly demonstrated that the homolog-scanning mutations, including those that destroy ligand binding, did not significantly change the overall structure of the intact receptor (14,18,19). Therefore, the loss of function was attributed to perturbations in the ligand recognition site. Indeed, the ligand binding sites we identified earlier within the ␣ M I-domain fit very well with the data obtained from the crystal structure of the collagen-␣ 2 I-domain complex (the ␣ 2 I-domain is a homolog of the ␣ M I-domain) (21). Given such a success and the similarity between the I-domains of the integrin ␣ and ␤ subunits (20,22), we conducted similar homolog-scanning mutagenesis on the ␤ 2 I-domain by substituting the sequences of ␤ 2 with their homologous counterparts of ␤ 1 (13). A total of 17 homolog mutants were constructed for this study, and four of these mutations had significant effects on ligand binding, suggesting that the ligand binding site is composed of these four segments. This conclusion is supported by the following observations: 1) all negative ␣ M ␤ 2 mutants could be expressed well on the cell surface as correct heterodimers and were recognized by a panel of conformation-dependent mAbs (13); 2) the four defective mutants still possessed an intact and high affinity Ca 2ϩ -binding site located within the ␤ 2 subunit (Fig. 2); 3) one peptide (P6), located within the identified ligand binding site, could effectively compete with the intact ␣ M ␤ 2 receptor for ligand binding; and 4) when immobilized on the plastic surface, peptides P6 and P11 could directly bind the ␥-module and C3bi. Although peptides P1 and P17 did not exhibit detectable ligand binding activity, the negative results do not exclude a role for these peptides in ligand binding. The immobilized peptides may simply not adopt the appropriate conformation for recognition by the ligands. The ligand binding site we determined in this work agrees well with the observation by Goodman et al. (17) that residues Asp 232 and Glu 235 , which reside within the identified ligand binding site, are critical to Fg and C3bi binding by ␣ M ␤ 2 .
To locate these four functional segments within the ␤ 2 Idomain, we modeled its three-dimensional structure based on the recently published crystal coordinates of integrin ␣ V ␤ 3 (16). As shown in Fig. 5, the four segments (Arg 144 -Lys 148 , Gln 199 -Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 ) reside in the same region on the apex of the ␤ 2 I-domain, supported by the underlying ␤-strands. The proximity of these identified segments in the three-dimensional structure is consistent with a role of these sequences in ligand recognition. Furthermore, this identified binding site contains the conserved DXSXS sequence and the predicted Mg 2ϩ and Ca 2ϩ binding sites, which are essential to ␣ M ␤ 2 -ligand interactions (23). Based on the data from our earlier study (13) and from the recently published crystal structure of the RGD-␣ V ␤ 3 complex (16), the principle function of the DXSXS sequence is likely to provide a scaffold for high affinity binding of divalent cations to the ␤I-domain. These bound cations, in turn, help to maintain correct conformations of the ␤I-domain (13), and at the same time, they also contribute directly to ligand binding by contacting with the acidic residues of the ligand (16). Thus, the location of our identified ligand binding site agrees well with the critical roles of the DXSXS sequence and these divalent ions in ligand recognition.
Our results suggest that sequences Arg 144 -Lys 148 , Gln 199 - Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 contribute to the formation of ligand binding sites within the ␤ 2 I-domain. Interestingly, this identified ligand binding site resembles very much the ligand binding pocket within the ␣ M I-domain, in that both ligand binding pockets encompass a broad region within their respective I-domains, containing the conserved DXSXS sequence and one or more cation binding sites (Fig. 5) (18,19). In particular, we have reported that NIF (an ␣ M ␤ 2 -specific ligand) and C3bi recognize overlapping but nonidentical binding sites with the ␣ M I-domain (11). As shown in Fig. 5B, the region shared by these two ligands is composed of three segments (Pro 147 -Arg 152 , Pro 201 -Lys 217 , and Asp 248 -Arg 261 ), whereas residue Lys 245 is recognized differentially by C3bi and NIF (18,19). Although the Fg binding site within the ␣ M I-domain has not been fully delineated, our earlier study shows that it overlaps with the NIF and C3bi binding sites and contains yet another segment ( 281 RQELNTI) that resides outside the identified NIF and C3bi binding pockets (11), indicating that the Fg binding site occupies an even broader region than those of NIF and C3bi. Similar results were obtained for the ␤ 2 I-domain, where Fg and C3bi recognize an overlapping region (Gln 199 -Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 ) and a differentially recognized region (Arg 144 -Lys 148 ). Finally, the ligand binding site we identified here for the ␤ 2 I-domain agrees well with the recently determined ligand binding site within the ␤ 3 I-domain, containing residues Tyr 122 and 214 RNR 216 (16), which are equivalent to residues located within the ␤ 2 peptides P1 and P17, respectively. Interestingly, compared with the ␤ 3 I-domain, the ␤ 2 I-domain appears to utilize a much broader region for binding of its ligands, which may reflect the different requirements for recognition of protein versus peptide ligands or different mechanisms for ligand recognition by the ␤ 2 versus ␤ 3 integrins.
Our finding that both ␣ M and ␤ 2 are involved directly in ligand recognition also implies that the proper alignment of the two binding domains will be critical to the formation of a high affinity ligand binding site within the ␣ M ␤ 2 heterodimer. Therefore, the ␤ 2 subunit may also play a regulatory role in ligand binding. In support of such a notion, we have reported that recognition of Candida albicans by ␣ M ␤ 2 is mediated mainly by the ␣ M subunit, and the ␤ 2 subunit influences ligand binding by modulating the activity of ␣ M (24). Similarly, it was reported that the ␤ 2 subunit plays an indirect regulatory role in ligand binding by ␣ L ␤ 2 (25). In addition, it was proposed recently that the ␤ 2 I-domain may actually recognize the ␣ M subunit as a ligand and thereby controls ligand binding by ␣ M ␤ 2 (26). Altogether, these data suggest that the ␣ M and ␣ L I-domains provide major ligand contact sites, whereas the ␤ 2 subunit contributes directly and/or indirectly to ligand binding, depending on the nature of the individual ligands.
In summary, using a combination of different approaches, we have demonstrated that the ␤ 2 subunit contributes directly to ligand binding by ␣ M ␤ 2 . The recognition sites for the two representative ligands (Fg and C3bi) reside in a narrow region, composed of four segments, Arg 144 -Lys 148 , Gln 199 -Ala 203 , Leu 225 -Leu 230 , and Gly 305 -His 309 , on the apex of the ␤ 2 I-domain, with the first segment playing a more prominent role in Fg binding than in C3bi binding. The three potential cation binding sites (Mg 2ϩ and Ca 2ϩ ) within the ␤ 2 I-domain are located inside this identified region, suggesting that these three ions may contribute directly to ligand binding by the ␣ M ␤ 2 receptor. Since both the ␣ M and the ␤ 2 subunit contribute directly to ligand recognition, alignment of the two binding domains within the heterodimeric ␣ M ␤ 2 receptor will be critical to the formation of a proper ligand binding pocket, thus providing a potential mechanism for modulation of integrin activities. Given the similarity among all integrin ␤ subunits, our results may help us to understand the underlying mechanism of integrin-ligand interactions in general.